Annotation of mandoc/man_macro.c, Revision 1.99
1.99 ! schwarze 1: /* $Id: man_macro.c,v 1.98 2015/02/06 11:54:36 schwarze Exp $ */
1.1 kristaps 2: /*
1.65 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.93 schwarze 4: * Copyright (c) 2012, 2013, 2014, 2015 Ingo Schwarze <schwarze@openbsd.org>
1.79 schwarze 5: * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
1.14 kristaps 8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 10: *
1.14 kristaps 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 18: */
1.30 kristaps 19: #include "config.h"
1.88 schwarze 20:
21: #include <sys/types.h>
1.30 kristaps 22:
1.1 kristaps 23: #include <assert.h>
24: #include <ctype.h>
25: #include <stdlib.h>
26: #include <string.h>
27:
1.58 kristaps 28: #include "man.h"
1.46 kristaps 29: #include "mandoc.h"
1.57 kristaps 30: #include "libmandoc.h"
1.1 kristaps 31: #include "libman.h"
32:
1.34 kristaps 33: enum rew {
34: REW_REWIND,
35: REW_NOHALT,
1.40 kristaps 36: REW_HALT
1.34 kristaps 37: };
1.19 kristaps 38:
1.91 schwarze 39: static void blk_close(MACRO_PROT_ARGS);
40: static void blk_exp(MACRO_PROT_ARGS);
41: static void blk_imp(MACRO_PROT_ARGS);
42: static void in_line_eoln(MACRO_PROT_ARGS);
1.82 schwarze 43: static int man_args(struct man *, int,
1.60 kristaps 44: int *, char *, char **);
1.19 kristaps 45:
1.91 schwarze 46: static void rew_scope(enum man_type,
1.33 kristaps 47: struct man *, enum mant);
1.82 schwarze 48: static enum rew rew_dohalt(enum mant, enum man_type,
1.19 kristaps 49: const struct man_node *);
1.82 schwarze 50: static enum rew rew_block(enum mant, enum man_type,
1.21 kristaps 51: const struct man_node *);
1.19 kristaps 52:
53: const struct man_macro __man_macros[MAN_MAX] = {
1.31 kristaps 54: { in_line_eoln, MAN_NSCOPED }, /* br */
1.66 schwarze 55: { in_line_eoln, MAN_BSCOPE }, /* TH */
56: { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SH */
57: { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SS */
1.99 ! schwarze 58: { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* TP */
1.66 schwarze 59: { blk_imp, MAN_BSCOPE }, /* LP */
60: { blk_imp, MAN_BSCOPE }, /* PP */
61: { blk_imp, MAN_BSCOPE }, /* P */
62: { blk_imp, MAN_BSCOPE }, /* IP */
63: { blk_imp, MAN_BSCOPE }, /* HP */
1.90 schwarze 64: { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SM */
65: { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SB */
1.19 kristaps 66: { in_line_eoln, 0 }, /* BI */
67: { in_line_eoln, 0 }, /* IB */
68: { in_line_eoln, 0 }, /* BR */
69: { in_line_eoln, 0 }, /* RB */
1.90 schwarze 70: { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* R */
71: { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* B */
72: { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* I */
1.19 kristaps 73: { in_line_eoln, 0 }, /* IR */
74: { in_line_eoln, 0 }, /* RI */
1.31 kristaps 75: { in_line_eoln, MAN_NSCOPED }, /* sp */
1.66 schwarze 76: { in_line_eoln, MAN_BSCOPE }, /* nf */
77: { in_line_eoln, MAN_BSCOPE }, /* fi */
1.92 schwarze 78: { blk_close, MAN_BSCOPE }, /* RE */
1.99 ! schwarze 79: { blk_exp, MAN_BSCOPE }, /* RS */
1.23 kristaps 80: { in_line_eoln, 0 }, /* DT */
1.28 kristaps 81: { in_line_eoln, 0 }, /* UC */
1.29 kristaps 82: { in_line_eoln, 0 }, /* PD */
1.45 joerg 83: { in_line_eoln, 0 }, /* AT */
1.49 kristaps 84: { in_line_eoln, 0 }, /* in */
1.53 kristaps 85: { in_line_eoln, 0 }, /* ft */
1.71 kristaps 86: { in_line_eoln, 0 }, /* OP */
1.72 schwarze 87: { in_line_eoln, MAN_BSCOPE }, /* EX */
88: { in_line_eoln, MAN_BSCOPE }, /* EE */
1.99 ! schwarze 89: { blk_exp, MAN_BSCOPE }, /* UR */
1.92 schwarze 90: { blk_close, MAN_BSCOPE }, /* UE */
1.81 schwarze 91: { in_line_eoln, 0 }, /* ll */
1.19 kristaps 92: };
1.9 kristaps 93:
1.19 kristaps 94: const struct man_macro * const man_macros = __man_macros;
1.1 kristaps 95:
96:
1.91 schwarze 97: void
1.83 schwarze 98: man_unscope(struct man *man, const struct man_node *to)
1.1 kristaps 99: {
1.55 kristaps 100: struct man_node *n;
1.1 kristaps 101:
1.83 schwarze 102: to = to->parent;
103: n = man->last;
104: while (n != to) {
1.87 schwarze 105:
106: /* Reached the end of the document? */
107:
108: if (to == NULL && ! (n->flags & MAN_VALID)) {
109: if (man->flags & (MAN_BLINE | MAN_ELINE) &&
110: man_macros[n->tok].flags & MAN_SCOPED) {
111: mandoc_vmsg(MANDOCERR_BLK_LINE,
112: man->parse, n->line, n->pos,
113: "EOF breaks %s",
114: man_macronames[n->tok]);
115: if (man->flags & MAN_ELINE)
116: man->flags &= ~MAN_ELINE;
117: else {
118: assert(n->type == MAN_HEAD);
119: n = n->parent;
120: man->flags &= ~MAN_BLINE;
121: }
122: man->last = n;
123: n = n->parent;
124: man_node_delete(man, man->last);
125: continue;
126: }
127: if (n->type == MAN_BLOCK &&
1.99 ! schwarze 128: man_macros[n->tok].fp == blk_exp)
1.87 schwarze 129: mandoc_msg(MANDOCERR_BLK_NOEND,
130: man->parse, n->line, n->pos,
131: man_macronames[n->tok]);
132: }
133:
1.55 kristaps 134: /*
1.83 schwarze 135: * We might delete the man->last node
136: * in the post-validation phase.
137: * Save a pointer to the parent such that
138: * we know where to continue the iteration.
1.55 kristaps 139: */
1.89 schwarze 140:
1.83 schwarze 141: man->last = n;
142: n = n->parent;
1.91 schwarze 143: man_valid_post(man);
1.19 kristaps 144: }
1.89 schwarze 145:
146: /*
147: * If we ended up at the parent of the node we were
148: * supposed to rewind to, that means the target node
149: * got deleted, so add the next node we parse as a child
150: * of the parent instead of as a sibling of the target.
151: */
152:
153: man->next = (man->last == to) ?
154: MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
1.19 kristaps 155: }
1.1 kristaps 156:
1.34 kristaps 157: static enum rew
1.33 kristaps 158: rew_block(enum mant ntok, enum man_type type, const struct man_node *n)
1.21 kristaps 159: {
160:
1.91 schwarze 161: if (type == MAN_BLOCK && ntok == n->parent->tok &&
162: n->parent->type == MAN_BODY)
1.21 kristaps 163: return(REW_REWIND);
164: return(ntok == n->tok ? REW_HALT : REW_NOHALT);
165: }
166:
1.19 kristaps 167: /*
168: * There are three scope levels: scoped to the root (all), scoped to the
169: * section (all less sections), and scoped to subsections (all less
170: * sections and subsections).
171: */
1.82 schwarze 172: static enum rew
1.33 kristaps 173: rew_dohalt(enum mant tok, enum man_type type, const struct man_node *n)
1.19 kristaps 174: {
1.34 kristaps 175: enum rew c;
1.1 kristaps 176:
1.38 kristaps 177: /* We cannot progress beyond the root ever. */
1.19 kristaps 178: if (MAN_ROOT == n->type)
179: return(REW_HALT);
1.38 kristaps 180:
1.19 kristaps 181: assert(n->parent);
1.38 kristaps 182:
183: /* Normal nodes shouldn't go to the level of the root. */
1.19 kristaps 184: if (MAN_ROOT == n->parent->type)
185: return(REW_REWIND);
1.38 kristaps 186:
187: /* Already-validated nodes should be closed out. */
1.19 kristaps 188: if (MAN_VALID & n->flags)
189: return(REW_NOHALT);
190:
1.38 kristaps 191: /* First: rewind to ourselves. */
1.73 schwarze 192: if (type == n->type && tok == n->tok) {
1.99 ! schwarze 193: if (man_macros[n->tok].fp == blk_exp)
1.73 schwarze 194: return(REW_HALT);
195: else
196: return(REW_REWIND);
197: }
1.21 kristaps 198:
1.82 schwarze 199: /*
1.38 kristaps 200: * Next follow the implicit scope-smashings as defined by man.7:
201: * section, sub-section, etc.
202: */
203:
1.19 kristaps 204: switch (tok) {
1.82 schwarze 205: case MAN_SH:
1.19 kristaps 206: break;
1.82 schwarze 207: case MAN_SS:
1.20 kristaps 208: /* Rewind to a section, if a block. */
1.21 kristaps 209: if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
210: return(c);
211: break;
1.82 schwarze 212: case MAN_RS:
1.73 schwarze 213: /* Preserve empty paragraphs before RS. */
214: if (0 == n->nchild && (MAN_P == n->tok ||
215: MAN_PP == n->tok || MAN_LP == n->tok))
216: return(REW_HALT);
1.21 kristaps 217: /* Rewind to a subsection, if a block. */
218: if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
219: return(c);
220: /* Rewind to a section, if a block. */
221: if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
222: return(c);
1.19 kristaps 223: break;
224: default:
1.21 kristaps 225: /* Rewind to an offsetter, if a block. */
226: if (REW_NOHALT != (c = rew_block(MAN_RS, type, n)))
227: return(c);
1.20 kristaps 228: /* Rewind to a subsection, if a block. */
1.21 kristaps 229: if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
230: return(c);
1.20 kristaps 231: /* Rewind to a section, if a block. */
1.21 kristaps 232: if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
233: return(c);
1.19 kristaps 234: break;
1.3 kristaps 235: }
1.1 kristaps 236:
1.19 kristaps 237: return(REW_NOHALT);
238: }
1.9 kristaps 239:
1.19 kristaps 240: /*
241: * Rewinding entails ascending the parse tree until a coherent point,
242: * for example, the `SH' macro will close out any intervening `SS'
243: * scopes. When a scope is closed, it must be validated and actioned.
244: */
1.91 schwarze 245: static void
1.75 schwarze 246: rew_scope(enum man_type type, struct man *man, enum mant tok)
1.19 kristaps 247: {
248: struct man_node *n;
1.34 kristaps 249: enum rew c;
1.7 kristaps 250:
1.75 schwarze 251: for (n = man->last; n; n = n->parent) {
1.82 schwarze 252: /*
1.19 kristaps 253: * Whether we should stop immediately (REW_HALT), stop
254: * and rewind until this point (REW_REWIND), or keep
255: * rewinding (REW_NOHALT).
256: */
257: c = rew_dohalt(tok, type, n);
258: if (REW_HALT == c)
1.91 schwarze 259: return;
1.19 kristaps 260: if (REW_REWIND == c)
1.7 kristaps 261: break;
1.6 kristaps 262: }
1.1 kristaps 263:
1.82 schwarze 264: /*
1.38 kristaps 265: * Rewind until the current point. Warn if we're a roff
266: * instruction that's mowing over explicit scopes.
267: */
1.19 kristaps 268:
1.91 schwarze 269: man_unscope(man, n);
1.19 kristaps 270: }
271:
1.6 kristaps 272:
1.36 kristaps 273: /*
274: * Close out a generic explicit macro.
275: */
1.91 schwarze 276: void
1.21 kristaps 277: blk_close(MACRO_PROT_ARGS)
278: {
1.82 schwarze 279: enum mant ntok;
1.21 kristaps 280: const struct man_node *nn;
1.93 schwarze 281: char *p;
282: int nrew, target;
1.21 kristaps 283:
1.93 schwarze 284: nrew = 1;
1.21 kristaps 285: switch (tok) {
1.82 schwarze 286: case MAN_RE:
1.21 kristaps 287: ntok = MAN_RS;
1.93 schwarze 288: if ( ! man_args(man, line, pos, buf, &p))
289: break;
290: for (nn = man->last->parent; nn; nn = nn->parent)
291: if (nn->tok == ntok && nn->type == MAN_BLOCK)
292: nrew++;
293: target = strtol(p, &p, 10);
294: if (*p != '\0')
295: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
296: line, p - buf, "RE ... %s", p);
297: if (target == 0)
298: target = 1;
299: nrew -= target;
300: if (nrew < 1) {
301: mandoc_vmsg(MANDOCERR_RE_NOTOPEN, man->parse,
302: line, ppos, "RE %d", target);
303: return;
304: }
1.76 schwarze 305: break;
1.82 schwarze 306: case MAN_UE:
1.76 schwarze 307: ntok = MAN_UR;
1.21 kristaps 308: break;
309: default:
310: abort();
311: /* NOTREACHED */
312: }
313:
1.75 schwarze 314: for (nn = man->last->parent; nn; nn = nn->parent)
1.93 schwarze 315: if (nn->tok == ntok && nn->type == MAN_BLOCK && ! --nrew)
1.21 kristaps 316: break;
317:
1.91 schwarze 318: if (nn == NULL) {
1.84 schwarze 319: mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
320: line, ppos, man_macronames[tok]);
1.91 schwarze 321: rew_scope(MAN_BLOCK, man, MAN_PP);
1.95 schwarze 322: } else {
323: line = man->last->line;
324: ppos = man->last->pos;
325: ntok = man->last->tok;
1.83 schwarze 326: man_unscope(man, nn);
1.95 schwarze 327:
328: /* Move a trailing paragraph behind the block. */
329:
330: if (ntok == MAN_LP || ntok == MAN_PP || ntok == MAN_P) {
331: *pos = strlen(buf);
332: blk_imp(man, ntok, line, ppos, pos, buf);
333: }
334: }
1.21 kristaps 335: }
336:
1.91 schwarze 337: void
1.35 kristaps 338: blk_exp(MACRO_PROT_ARGS)
339: {
1.98 schwarze 340: struct man_node *head;
341: char *p;
1.59 kristaps 342: int la;
1.35 kristaps 343:
1.91 schwarze 344: rew_scope(MAN_BLOCK, man, tok);
345: man_block_alloc(man, line, ppos, tok);
346: man_head_alloc(man, line, ppos, tok);
1.98 schwarze 347: head = man->last;
1.35 kristaps 348:
1.98 schwarze 349: la = *pos;
350: if (man_args(man, line, pos, buf, &p))
1.91 schwarze 351: man_word_alloc(man, line, la, p);
1.35 kristaps 352:
1.98 schwarze 353: if (buf[*pos] != '\0')
354: mandoc_vmsg(MANDOCERR_ARG_EXCESS,
355: man->parse, line, *pos, "%s ... %s",
356: man_macronames[tok], buf + *pos);
1.73 schwarze 357:
1.98 schwarze 358: man_unscope(man, head);
1.91 schwarze 359: man_body_alloc(man, line, ppos, tok);
1.35 kristaps 360: }
361:
1.19 kristaps 362: /*
363: * Parse an implicit-block macro. These contain a MAN_HEAD and a
364: * MAN_BODY contained within a MAN_BLOCK. Rules for closing out other
365: * scopes, such as `SH' closing out an `SS', are defined in the rew
366: * routines.
367: */
1.91 schwarze 368: void
1.19 kristaps 369: blk_imp(MACRO_PROT_ARGS)
370: {
1.59 kristaps 371: int la;
1.19 kristaps 372: char *p;
1.25 kristaps 373: struct man_node *n;
1.19 kristaps 374:
1.91 schwarze 375: rew_scope(MAN_BODY, man, tok);
376: rew_scope(MAN_BLOCK, man, tok);
377: man_block_alloc(man, line, ppos, tok);
378: man_head_alloc(man, line, ppos, tok);
1.75 schwarze 379: n = man->last;
1.25 kristaps 380:
1.19 kristaps 381: /* Add line arguments. */
1.3 kristaps 382:
1.19 kristaps 383: for (;;) {
384: la = *pos;
1.75 schwarze 385: if ( ! man_args(man, line, pos, buf, &p))
1.19 kristaps 386: break;
1.91 schwarze 387: man_word_alloc(man, line, la, p);
1.6 kristaps 388: }
389:
1.99 ! schwarze 390: /*
! 391: * For macros having optional next-line scope,
! 392: * keep the head open if there were no arguments.
! 393: * For `TP', always keep the head open.
! 394: */
1.19 kristaps 395:
1.99 ! schwarze 396: if (man_macros[tok].flags & MAN_SCOPED &&
! 397: (tok == MAN_TP || n == man->last)) {
! 398: man->flags |= MAN_BLINE;
! 399: return;
1.27 kristaps 400: }
1.99 ! schwarze 401:
! 402: /* Close out the head and open the body. */
! 403:
1.91 schwarze 404: rew_scope(MAN_HEAD, man, tok);
405: man_body_alloc(man, line, ppos, tok);
1.4 kristaps 406: }
407:
1.91 schwarze 408: void
1.19 kristaps 409: in_line_eoln(MACRO_PROT_ARGS)
1.3 kristaps 410: {
1.59 kristaps 411: int la;
1.19 kristaps 412: char *p;
413: struct man_node *n;
1.3 kristaps 414:
1.91 schwarze 415: man_elem_alloc(man, line, ppos, tok);
1.75 schwarze 416: n = man->last;
1.3 kristaps 417:
1.19 kristaps 418: for (;;) {
1.96 schwarze 419: if (buf[*pos] != '\0' && (tok == MAN_br ||
420: tok == MAN_fi || tok == MAN_nf)) {
421: mandoc_vmsg(MANDOCERR_ARG_SKIP,
422: man->parse, line, *pos, "%s %s",
1.97 schwarze 423: man_macronames[tok], buf + *pos);
424: break;
425: }
426: if (buf[*pos] != '\0' && man->last != n &&
427: (tok == MAN_PD || tok == MAN_ft || tok == MAN_sp)) {
428: mandoc_vmsg(MANDOCERR_ARG_EXCESS,
429: man->parse, line, *pos, "%s ... %s",
1.96 schwarze 430: man_macronames[tok], buf + *pos);
431: break;
432: }
1.19 kristaps 433: la = *pos;
1.75 schwarze 434: if ( ! man_args(man, line, pos, buf, &p))
1.19 kristaps 435: break;
1.90 schwarze 436: if (man_macros[tok].flags & MAN_JOIN &&
437: man->last->type == MAN_TEXT)
438: man_word_append(man, p);
1.91 schwarze 439: else
440: man_word_alloc(man, line, la, p);
1.19 kristaps 441: }
1.78 schwarze 442:
443: /*
444: * Append MAN_EOS in case the last snipped argument
445: * ends with a dot, e.g. `.IR syslog (3).'
446: */
447:
448: if (n != man->last &&
1.80 schwarze 449: mandoc_eos(man->last->string, strlen(man->last->string)))
1.78 schwarze 450: man->last->flags |= MAN_EOS;
1.3 kristaps 451:
1.31 kristaps 452: /*
453: * If no arguments are specified and this is MAN_SCOPED (i.e.,
454: * next-line scoped), then set our mode to indicate that we're
455: * waiting for terms to load into our context.
456: */
457:
1.91 schwarze 458: if (n == man->last && man_macros[tok].flags & MAN_SCOPED) {
459: assert( ! (man_macros[tok].flags & MAN_NSCOPED));
1.75 schwarze 460: man->flags |= MAN_ELINE;
1.91 schwarze 461: return;
1.31 kristaps 462: }
1.64 kristaps 463:
1.91 schwarze 464: assert(man->last->type != MAN_ROOT);
1.75 schwarze 465: man->next = MAN_NEXT_SIBLING;
1.82 schwarze 466:
1.19 kristaps 467: /*
1.31 kristaps 468: * Rewind our element scope. Note that when TH is pruned, we'll
469: * be back at the root, so make sure that we don't clobber as
470: * its sibling.
1.19 kristaps 471: */
1.3 kristaps 472:
1.75 schwarze 473: for ( ; man->last; man->last = man->last->parent) {
474: if (man->last == n)
1.19 kristaps 475: break;
1.75 schwarze 476: if (man->last->type == MAN_ROOT)
1.19 kristaps 477: break;
1.91 schwarze 478: man_valid_post(man);
1.19 kristaps 479: }
1.3 kristaps 480:
1.75 schwarze 481: assert(man->last);
1.3 kristaps 482:
483: /*
1.82 schwarze 484: * Same here regarding whether we're back at the root.
1.3 kristaps 485: */
486:
1.91 schwarze 487: if (man->last->type != MAN_ROOT)
488: man_valid_post(man);
1.19 kristaps 489: }
1.3 kristaps 490:
491:
1.91 schwarze 492: void
1.75 schwarze 493: man_macroend(struct man *man)
1.19 kristaps 494: {
1.22 kristaps 495:
1.91 schwarze 496: man_unscope(man, man->first);
1.19 kristaps 497: }
1.3 kristaps 498:
1.60 kristaps 499: static int
1.75 schwarze 500: man_args(struct man *man, int line, int *pos, char *buf, char **v)
1.60 kristaps 501: {
502: char *start;
503:
504: assert(*pos);
505: *v = start = buf + *pos;
506: assert(' ' != *start);
507:
508: if ('\0' == *start)
509: return(0);
510:
1.75 schwarze 511: *v = mandoc_getarg(man->parse, v, line, pos);
1.60 kristaps 512: return(1);
513: }
CVSweb